Roland Dreier wrote:
I finally looked this over.
First, this should be two patches: making srp_sg_tablesize tunable
should be a separate change from making it possible to specify
max_cmd_per_lun for a target.
OK, I'll break it to two patches
The srp_sg_tablesize change makes the default number of SG entries
quite a bit larger than it is now, which makes the default max IU
length much bigger. Is this justified? What workload creates such
huge SG lists?
With semi-realistic benchmark xdd, orion I'm seeing better
number with this default value
I think that we can reduce the default value for
srp_sg_tablesize and up to users to bump it up by overriding
when loading up the module.
For the cmd_per_lun change, shouldn't the line
+ target->scsi_host->cmd_per_lun = token;
be something like
target->scsi_host->cmd_per_lun = min(token,
SRP_SQ_SIZE);
otherwise it's too easy to overflow a send queue by mistake.
Yes. I'll fix it
Vu
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
